A remote object storage file system for Mahara. Intended to provide a plug-in that can be installed and configured to work with any supported remote object storage solution. Currently supports both AWS S3 and Azure Blob Storage.
This plug-in requires either mahara-module_aws or mahara-module_azure to function.
- Use cases
- Installation
- Currently supported object stores
- Mahara configuration
- Backporting
- Crafted by Catalyst IT
- Contributing and support
There are a number of different ways you can use this plug in. See Recommended use case settings for recommended settings for each one.
Disk can be expensive, so a simple use case is we simply want to move some of the largest and oldest files off local disk to somewhere cheaper. But we still want the convenience and performance of having the majority of files local, especially if you are hosting on-prem where the latency or bandwidth to the remote filesystem may not be great.
Many of our clients have multiple mahara instances, and there is much duplicated content across instances. By pointing multiple maharas at the same remote filesystem, and not allowing deletes, then large amounts of content can be de-duplicated.
Some of our clients maharas are truly massive. We also have multiple environments for various types of testing, and often have ad hoc environments created on demand. Not only do we not want to have to store duplicated files, but we also want refreshing data to new environments to be as fast as possible.
Using this plugin we can configure production to have full read write to the remote filesystem and store the vast bulk of content remotely. In this setup the latency and bandwidth isn't an issue as they are colocated. The local filedir on disk would only consist of small or fast churning files. A refresh of the production data back to a staging environment can be much quicker now as we skip the sitedir clone completely and stage is simple configured with readonly access to the production filesystem. Any files it creates would only be written to it's local filesystem which can then be discarded when next refreshed.
- If not on Mahara 18.04, backport the file system API. See Backporting
- Setup your remote object storage. See Remote object storage setup
- Clone this repository into module/objectfs
- Clone either mahara-module_aws into module/aws or mahara-module_azure into module/azure
- Install the plugins through the mahara GUI.
- Configure the plugin. See Mahara configuration
- Place one of the following in htdocs/config.php depending on your choice of remote storage (Azure or s3):
For AWS S3:
$cfg->externalfilesystem = array(
"includefilepath" => "module/objectfs/classes/s3_file_system.php",
"class" => "module_objectfs\\s3_file_system"
);
For Azure:
$cfg->externalfilesystem = array(
"includefilepath" => "module/objectfs/classes/azure_file_system.php",
"class" => "module_objectfs\\azure_file_system"
);
There is support for more object stores planed, in particular enabling Openstack deployments.
Amazon S3 bucket setup
- Create an Amazon S3 bucket.
- The AWS Users access policy should mirror the policy listed below.
- Replace 'bucketname' with the name of your S3 bucket.
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": ["s3:ListBucket"],
"Resource": ["arn:aws:s3:::bucketname"]
},
{
"Effect": "Allow",
"Action": [
"s3:PutObject",
"s3:GetObject",
"s3:DeleteObject"
],
"Resource": ["arn:aws:s3:::bucketname/*"]
}
]
}Azure Storage container guide with the CLI
It is possible to install the Azure CLI locally to administer the storage account. The Azure CLI can be obtained here.
Visit the Online Azure Portal or use the Azure CLI to obtain the storage account keys. These keys are used to setup the container, configure an access policy and acquire a Shared Access Signature that has Read and Write capabilities on the container.
It will be assumed at this point that a resource group and blob storage account exists.
- Obtain the account keys.
az login
az storage account keys list \
--resource-group <resource_group_name> \
--account-name <storage_account_name>
- Create a private container in a storage account.
az storage container create \
--name <container_name> \
--account-name <storage_account_name> \
--account-key <storage_account_key> \
--public-access off \
--fail-on-exist
- Create a stored access policy on the containing object.
az storage container policy create \
--account-name <storage_account_name> \
--account-key <storage_account_key> \
--container-name <container_name> \
--name <policy_name> \
--start <YYYY-MM-DD> \
--expiry <YYYY-MM-DD> \
--permissions rw
# Start and Expiry are optional arguments.
- Generates a shared access signature for the container. This is associated with a policy.
az storage container generate-sas \
--account-name <storage_account_name> \
--account-key <storage_account_key> \
--name <container_name> \
--policy <policy_name> \
--output tsv
- If you wish to revoke access to the container, remove the policy which will invalidate the SAS.
az storage container policy delete \
--account-name <storage_account_name> \
--account-key <storage_account_key> \
--container-name <container_name>
--name <policy_name>
Go to Administration -> Extensions -> objectfs. Descriptions for the various settings are as follows:
- Enable file transfer tasks: Enable or disable the object file system tasks which move files between the filedir and remote object storage.
- Maximum task runtime: Background tasks handle the transfer of objects to and from remote object storage. This setting controls the maximum runtime for all object transfer related tasks.
- Prefer remote objects: If a file is stored both locally and in remote object storage, read from remote. This is setting is mainly for testing purposes and introduces overhead to check the location.
These settings control the movement of files to and from object storage.
- Minimum size threshold (KB): Minimum size threshold for transferring objects to remote object storage. If objects are over this size they will be transfered.
- Minimum age: Minimum age that a object must exist on the local filedir before it will be considered for transfer.
- Delete local objects: Delete local objects once they are in remote object storage after the consistency delay.
- Consistency delay: How long an object must have existed after being transfered to remote object storage before they are a candidate for deletion locally.
This settings controls which object storage type to use, depending on installed modules the current choices are \module_objectfs\s3_filesystem
and \module_objectfs\azure_file_system
Azure specific settings
- Account name: The name of the storage account created in Azure
- Container name: The name of the container in Azure that will store the blobs
- Share access signature: The Shared Access Signature that you created for access to the container
S3 specific settings
- Key: AWS credential key
- Secret: AWS credential secret
- Bucket: S3 bucket name to store files in
- AWS region: AWS API endpoint region to use.
A major change in this release impacts all files stored in the module_objectfs_objects table. The ID stored in this table have changed from referencing the fileid to referencing the artefactid.
If you have moved files to S3 and have not deleted local files, it is recommended to delete all records in the module_objectfs_objects and initiate the upload again.
If you have moved files to S3 and have deleted local files, a script will need to be written to update the IDs from fileid to artefactid based on the file hash. This script has not been written in this release. We welcome merge requests to implement this change.
If you are on an older mahara then you can backport the necessary API's in order to support this plugin. Use with caution!
Thanks to Microsoft for sponsoring the Azure Storage implementation.
This plugin was developed by Catalyst IT Australia:
Issues, and pull requests using github are welcome and encouraged!
https://github.com/catalyst/mahara-module_objectfs/issues
If you would like commercial support or would like to sponsor additional improvements to this plugin please contact us: